I'd like you to give me two examples of situations in which you think a higher alpha threshold would be okay. (that is, alpha greater than .05).
Give me two examples of situations in which you think a lower alpha threshold would be necessary.
Assignment 11:
Give me four examples of errors.
For each example, explain what at Type I error would be and what a Type II error would be.
These examples can be anything you like – medical tests, legal tests, mechanics, or whatever you like! Be creative, have a little “fun” with it.
Though, to be honest, this is never a course that will have the reputation of being fun. :)
Assignment 12:
Imagine I ran an experiment with my class. I assign half of the class to submit detailed notes based on the textbook, while the other half does not have to do so. I find at the end of the class that those who turned in notes achieved higher grades, and that this difference was statistically significant.
What threshold would you draw for practical significance here? That is, how much higher would the note-taking students have to score for me to consider assigning this method for all future students in the course? Why? I want a brief paragraph's explanation.
Note: I have not run this study, nor do I actually plan to make this change. I am merely asking for your opinion about this, from a scientist-statistician perspective. :)